Category Archives: General statistics

Bayesian ANOVA: Powerful inference with within-group sample size of 1

By | March 9, 2017

1 Objective 2 The data 3 Fixed-effects ANOVA in JAGS 4 Relaxing the assumption of constant variance 5 Conclusion This post is inspired by a question by Dylan Craven that he raised during my Bayesian stats course. 1 Objective My aim here is to demonstrate that, in Bayesian setting, one can make powerful inference about… Read More »

Kéry & Royle have a new book on hierarchical modeling in ecology. It's good

By | January 7, 2016

Marc Kéry's books are as important for learning (and teaching) hierarchical modeling as Crawley's The R Book is for learning R. I hold Kéry's Introduction to WinBUGS high for the uncompromising didactic clarity. J. Andrew Royle is one of the founding minds (with James Nichols and Darryl MacKenzie) of the so called occupancy modeling, and… Read More »

Bayesian PCA

By | January 5, 2015

Authors: Jan Smycka, Petr Keil This post introduces experimental R package bPCA which we developed with Jan Smycka, who actually came with the idea. We do not guarantee the very idea to be correct and there certainly are bugs – we invite anyone to show us wrong, or to contribute. Rationale of bPCA Here is… Read More »

Is my brilliant idea any good? I am not sure, so I've pre-printed it on PeerJ

By | July 24, 2014

As a scientist, what should I do when I encounter a seemingly fundamental problem that also seems strangely unfamiliar? Is it unfamiliar because I am up to something really new, or am I re-discovering something that has been around for centuries, and I have just missed it? This is a short story about an exploration… Read More »

Poisson regression fitted by glm(), maximum likelihood, and MCMC

By | October 30, 2013

The goal of this post is to demonstrate how a simple statistical model (Poisson log-linear regression) can be fitted using three different approaches. I want to demonstrate that both frequentists and Bayesians use the same models, and that it is the fitting procedure and the inference that differs. This is also for those who understand… Read More »

Do simple models lead to generality in ecology? Opinion of a simpleton

By | September 25, 2013

Evans et al. have a paper in Trends in Ecology and Evolution with this abstract: Modellers of biological, ecological, and environmental systems cannot take for granted the maxim ‘simple means general means good’. We argue here that viewing simple models as the main way to achieve generality may be an obstacle to the progress of… Read More »

The joy and martyrdom of trying to be a Bayesian

By | August 30, 2013

Some of my fellow scientists have it easy. They use predefined methods like linear regression and ANOVA to test simple hypotheses; they live in the innocent world of bivariate plots and lm(). Sometimes they notice that the data have odd histograms and they use glm(). The more educated ones use generalized linear mixed effect models.… Read More »

Seeing Pierre Legendre

By | June 10, 2013

As suggested by his name, the guy is a legend. One of the most cited authors in ecology, I have him in (almost) the same league with James H. Brown, sir Robert M. May or Stephen P. Hubbell. Legendre is not famous for creating a revolutionary ecological theory and he does not stand out as… Read More »

AIC & BIC vs. Crossvalidation

By | May 5, 2013

Model selection is a process of seeking the model in a set of candidate models that gives the best balance between model fit and complexity (Burnham & Anderson 2002). I have always used AIC for that. But you can also do that by crossvalidation. Specifically, Stone (1977) showed that the AIC and leave-one out crossvalidation… Read More »

On ensemble forecasting

By | April 27, 2013

Yesterday, professor Ronald Smith gave a talk at Yale about how to predict future climate. One of his central subjects was ensemble forecasting. Here I give it a bit of a dissection. Climatologists and ecologists do "predictive models". Once they have the model, they use it to predict the future, e.g.: How will global temperature… Read More »

Not all proportion data are binomial outcomes

By | March 24, 2013

It really is trivial. Not every proportion is frequency. There are things that have values  bounded between 0 and 1 and yet they are neither probabilities, nor frequencies. Why do I even bother to write this? Because some kinds of proportions should be treated as unbounded continuous variables, and should be analyzed using appropriate statistical… Read More »

Predictors, responses and residuals: What really needs to be normally distributed?

By | February 18, 2013

Introduction Many scientists are concerned about normality or non-normality of variables in statistical analyses. The following and similar sentiments are often expressed, published or taught: "If you want to do statistics, then everything needs to be normally distributed." "We normalized our data in order to meet the assumption of normality." "We log-transformed our data as… Read More »